Brazil Pollsters Criticize TSE Plan to Reward Election Survey Accuracy

Brazil's main association of polling companies has criticized a proposal by the Superior Electoral Court (TSE), Brazil's top electoral court, to create an award for pollsters whose surveys come closest to official election results. The Brazilian Association of Research Companies (Abep) said the idea confuses scientific measurement with prediction, based on single-source reporting from O Globo.

The proposal, presented on Tuesday, July 14, by TSE President Kassio Nunes Marques, would create an "Electoral Accuracy Seal" for companies whose projections show the strongest alignment with the official vote count in the 2026 elections. The draft was discussed with representatives of 16 polling firms.

The Pollsters' Objection

Abep, whose members include Datafolha, Quaest and Ipsos, said polls measure voter intention at the moment interviews are conducted. They are not promises of the final result, the association said, because voters can change their minds, abstain or alter their behavior before election day.

"Requiring a poll to 'get' the result right is confusing science with a crystal ball," Abep said in a statement quoted by O Globo.

The association argued that the proposed seal could create what it called a perverse incentive. Less rigorous firms, it said, could follow the findings of more established institutes and adjust their numbers toward a perceived consensus in order to increase their chances of winning recognition.

Abep also said the quality of a survey should be assessed through methodology, sample design, transparency, fieldwork and adherence to scientific standards, not only by how close a pre-election snapshot came to a later result.

What the TSE Proposed

According to O Globo, the draft would apply only in general election years. The TSE would recognize national presidential polls, while regional electoral courts would be responsible for gubernatorial polling.

The proposal would consider only surveys properly registered in PesqEle, the electoral polling registry system, and actually released to the public. It foresees two categories: exit polls and surveys published during the seven days before voting.

The draft also lists grounds for exclusion, including final convictions for fraudulent polling, serious irregularities that lead to suspension or invalidation of a survey, or failure to meet technical criteria to be defined in future regulation.

Nunes Marques told participants that the text circulated on Tuesday was an initial proposal and that the court would continue receiving suggestions until Friday. O Globo reported that Datafolha, Quaest, Ipec, Ipsos-Ipespe and MDA opposed the seal during the meeting.

A Wider Debate

The discussion follows a dispute over a suspended AtlasIntel poll about the political impact of the Banco Master case on Senator Flávio Bolsonaro, a member of the Liberal Party (PL) and son of former President Jair Bolsonaro. In late May, Nunes Marques halted publication of the survey after a complaint from the PL, finding that some questions could induce responses by directly linking the case to the senator.

That decision triggered criticism from AtlasIntel and opened a broader debate among specialists and polling firms over how far the Electoral Court should go in supervising surveys.

The TSE is also considering a wider rule on transparency, registration and disclosure of election polls. One possible measure, according to O Globo, would require institutes to describe their methodology in greater detail, including whether questions were spontaneous or prompted and what criteria were used in the questionnaire.

Nunes Marques said the court does not intend to discuss specific poll results or interfere with the technical autonomy of research companies. He framed the meeting as an effort to build permanent cooperation, increase public confidence and give companies greater legal certainty.

For Abep, however, the concern is that an official accuracy prize would make the Electoral Court an arbiter of research quality. The association said such initiatives should be developed with the scientific community and polling institutes so they do not weaken the methodological rigor that serious surveys require.